RotoWire Partners

East Coast Offense: Predictions vs. Projections

Chris Liss

Chris Liss

Chris Liss is RotoWire's Managing Editor and Host of RotoWIre Fantasy Sports Today on Sirius XM radio.

Projections vs. Predictions

We raised this topic last week on the Sirius XM show, and it got a good response. The question was whether we want our fantasy pundits giving us projections, i.e., probability-based expected-return answers, or specific predictions of how things will actually go. For example, if you ask whether to start Cam Newton or Andy Dalton this week, I could answer: "I give Newton a 60 percent chance to outperform Dalton." Or I could just say: "Go Newton, I expect him to have a monster game." Which answer is preferable?

On its face, the easy answer is the former give us the odds, and let the chips fall. Unless your favorite pundit has psychic powers, the best he can do for you is tell you what the past data indicates and provide you with an accurate reading of what the most likely outcomes are. In the short run, there's no guarantee of being right, but over time, going with better probabilities is always going to yield better results.

But a flaw in this kind of probabilistic framing is that it's not falsifiable, i.e., no matter what happens, you can never be wrong because there was some possibility that the underdog outperformed the favorite. If Dalton in the above example crushes Newton, I can always say: "I said there was a 40 percent chance of that happening." Even if I said Player X was a 90/10 favorite to outperform Player Y, I can say the 10 percent happened to come through. After all, 10 percent, while unlikely, doesn't come close to passing the Airplane Test (percentage of crashing low enough to board).

So the probabilistic pundit can always avoid accountability by hiding behind the possibility that the less likely outcome obtained. When the Giants played the undefeated Patriots in Super Bowl XLII, one web site said it was "the biggest mismatch in Super Bowl history". When the Giants won that Super Bowl, they claimed the Giants were lucky.

It's possible the Giants were in fact lucky, and SB XLII was the one game out of 10 the Patriots would have lost should those teams have played 10 straight weeks (though it certainly didn't look like that while watching it). But it's also possible the numbers used to calculate the odds before the game were wrong. When faced with difficult new data, one can always try to shoehorn what you see into one's existing model, however ill the fit. Or dismiss it due to variance, i.e., "luck." Certainly that's far easier than re-designing the model to account for all of the observed phenomena.

So would we rather our pundits just make judgment calls? Last week on Fantasy Football Live, when asked whether Darren McFadden would go over or under 109.5 combined yards against the Falcons, I said: "Double." I thought McFadden would go off and made that call. The merit of this is that I'm clearly accountable it was wrong. I can't spin my way out of it by saying, "there's of course a chance he doesn't have a big game," especially when he stayed healthy, got a big workload and simply didn't have room to run. Moreover, the predicting pundit takes on the problem of variance rather than just throwing up his hands. The variance-neutral expectation so rarely obtains that even if one is right on average with a probabilistic approach, he's not usually correct in specific instances. It's like the old joke about two statisticians going hunting. One shoots five feet to the right of the deer, the other five feet to the left. Both chime in unison: "Got him!"

But the obvious downside of making that kind of call is it often doesn't have a basis in reality. My imagination of how the game will go is just that imagination an educated rendering of the future by an experienced and biased observer. No matter how keen one's instincts, no one can perfectly weight all of the prior match-ups and situations one's observed. Moreover, there's a good deal of luck in all professional sports, and without a rigorous model, it's hard not to be influenced by superficial results, many of which have little predicitive value going forward.

In the end, I think we need to incorporate both approaches. It's important to have a sound grasp of the causative factors in individual and team performance to know as best we can which are repeatable and which are aberrations. At the same time, probabilistic models are able to tell us only what might happen generally assuming the underlying conditions stay more or less the same. The general rules they discern almost certainly have exceptions, see the Giants' last two Super Bowl runs, and it would be better to carve out those exceptions and incorporate them into the overall system rather than dismissing them as "luck." Moreover, the systems are only valid so long as conditions don't change too much (See the Wall Street algorithms that worked great so long as housing prices never went down). But even if one acknowledges the proper limits of models and uses them with requisite restraint, there's still a role for one's personal observations, hunches and imagination in predicting the games. Not only are they falsifiable and therefore easy to grade but they take on the actual task at hand what will happen in this particular game, not hypothetically if these teams repeated the game a million times under the exact conditions. Finally, our observations allow us to identify situations where the statistical calculation might be missing something, or where the factors that go into the formula need to be weighted differently.

FG Kicking is Out of Control

There was a stretch a few years ago when kickers were so routinely missing chip shots with games on the line I wondered whether kicking had fallen off significantly. It turned out, the overall percentages were roughly in line with past seasons. But now that Matt Bryant hit a 55-yarder and Jay Feeley hit a 61-yarder (before missing a chip shot) Sunday, and kickers with huge legs from Greg Zuerlein to Sebastian Janikowski are populating the league, it deserves another look.

Through six games, the league average for field-goals made is a whopping 87.6 percent. Last year it was 82.9 percent, in 2010 it was 82.3 percent and in 2009 it was 81.3 percent.

This year, through six games teams have attempted an average of 1.7 (4.5 pace) FGs from 50-plus, making 1.1 (64.7%). Last year the average team through 16 games attempted 4.4 FG from 50-plus, making 2.8 of them (63.6%). In 2010, teams attempted 3.4 FGs from 50-plus, making 1.8 of them (53 percent).

It's fairly safe to say, kicking is peaking right now, both for accuracy and distance.

Things to Take Away from Week 6

I finally got DirecTV installed in the converted garage and hooked up to two 42-inch flatscreens. I leave one on the red-zone channel and another on whatever game I'm watching, flipping around during commercials. I'm sure many of you already have set-ups like this (and better), but I've always had either one screen at home with the package or the sports bar. Two big screens make a huge difference.

Seeing the Giants go into San Francisco and win by 23 was a nice way to break the new room in. The game wasn't as lopsided as the score Alex Smith's three picks set up some short fields, and the Niners defense held for the most part, though with a big lead, the Giants were content with field goals after a while.

But New York had six sacks, San Francisco had zero. And in the second half, Ahmad Bradshaw and David Wilson gashed the defense with tough runs, something no one does in that stadium. It was almost as if the Giants wanted to send a message not only do we have the better quarterback and the more explosive outside weapons, but we can beat you where you're strongest. It was like Rocky standing toe to toe with Clubber Lang and shoving his mohawked head: "Ain't so bad."

The Giants secondary is playing better. Prince Amukamara might be a difference maker opposite Corey Webster very likely an improvement over the departed Aaron Ross with whom they won a Super Bowl. The Giants pass rush got to Alex Smith, but it wasn't always right away. It seemed like at least a couple sacks were of the coverage variety. It might not be Revis Island or AlCROtraz, but maybe you don't want to get stranded in Port au Prince, either.

Ahmad Bradshaw now has 57 carries the last two games and seems to be holding up fine. If the Giants defense continues to play better, he can easily finish as a top-12 back. David Wilson looks like he's about to break a big play every time he touches the ball. For years, the Giants' return men were useless, but Wilson's already one of the best in the league. I don't see him getting a huge workload even if Bradshaw were to go down, however. In that case, he'd split carries with Andre Brown. But if Wilson ever averaged 8-10 rushes and 3-4 targets per game, he'd be useful.

The Ravens' defense is like the Steelers' not the same dominant unit you associate with the name brand. Maybe it's the absence of Terrell Suggs, but he's not coming back this year, so you'll have to adjust accordingly. They're 5-1, but they barely beat New England, Dallas and Kansas City, and the home game against Cleveland was competitive, too. Now Lardarius Webb and Ray Lewis are out for the year. The AFC North is turning into a wide-open finesse division while the NFC West home of the Bill Walsh Niners is the epicenter of smash-mouth.

Jason Garrett is a broken man. Scared to use his timeouts and get a more makeable FG, scared to go for it on 4th-and-5 from the Ravens' 35 earlier in the quarter. Of course, Dez Bryant dropped an easy two-point conversion that would have tied the game. It's always something with the Cowboys. They're the NFC version of the Chargers, and I'd be very surprised if Philip Rivers or Tony Romo ever win a Super Bowl despite being plenty good enough and having more than enough talent around them.

Jeff Fisher's Titans teams always seemed to be a tough out, and it's no different now that he's in St. Louis. Their only big loss was in Chicago, and even that game was close well into the fourth quarter. Neither Steven Jackson nor Daryl Richardson has a rushing touchdown this season, so Fisher hasn't abandoned the team's traditions altogether.

Say what you want about Michael Vick, but he puts up stats 300 yards, two passing TDs, 59 yards on the ground. Matt Stafford is still a volume guy, and he'll have plenty of passing yards on a per-game basis. But he's league average at best right now in real life, especially considering he has the best receiver in the league on his team.

Matt Ryan and the Falcons are awfully lucky. They completed one 20-yard pass after it bounced of a defender's helmet into Harry Douglas' hands while he was lying on the ground. They also got a pick six to take the lead, but gave it right back before Vanilla Ice heroically drove them into 55-yard FG range. If the season were to end today, Ryan would probably win the MVP vote, but it's telling that few if anyone would take him over Aaron Rodgers, Tom Brady, Peyton Manning, Eli Manning or Drew Brees. While the MVP isn't a career achievement award, you have to at least think the guy is top-five at his positon right now for him to deserve it.

Speaking of Aaron Rodgers, I told our FFL producer Alan Springer to start him over Peyton Manning, but he listened to the advice of others (who won't be named but who know who they are) and went with Manning. I like Peyton's match-up against the Chargers, but he's going to have a tough time beating 338 yards, six TDs and 17 rushing yards. Regrettably, a caller on that show asked Rodgers or Ryan, and I said Ryan! I knew it was wrong in my gut as I said it, but it was the "Rapid Response" segment, so there was no time to retract. I apologize to whoever that was. I also think the segment should be changed to "Rabid Reponse," but that's just me.

What happened to Brian Flatline? Did Jeff Fisher assign him "Finnegan's Wake?" (A novel that took James Joyce 17 years to write and, he expected, his readers that long to read)

I had A.J. Green as my No. 2 receiver heading into this year, and apparently I was off by one. Green just went 7-135-2 against Joe Haden who's now arguably the best corner in the league with Darrelle Revis out for the year. He's at least Calvin Johnson's equal in fantasy at this point.

Montario Hardesty passed the eye test when Trent Richardson went out with a rib injury. I'd be surprised if Richardson plays this week against the Colts. Brandon Weeden had another quietly competent game.

How did this Colts team beat the Packers last week? They made Shonn Greene look like Shaun Alexander in his prime.

Maybe the Chiefs should bring back Brodie Croyle instead of Brady Quinn.

The Bucs run defense seems fairly stout, holding Jamaal Charles to just 40 yards on 12 carries.

Carson Palmer is actually good, but he can't shake the pick-six habit. Denarius Moore is easily the team's best receiver when he's healthy. For some reason, Darren McFadden can't find a hole, but Mike Goodson can't find someone to tackle him. Goodson had two fewer yards on 25 fewer touches. Sebastian Janikowski booms 52-yard field-goals like they're extra points.

It was pretty frustrating to see Oakland blow that game as fading ATL was such a nice pot odds play in Survivor. Pretty much everyone else won this week except the Steelers whom I had in two of my three remaining pools.

Russell Wilson has played much better the last two games, but it was shocking to see the Pats let Sidney Rice beat them deep up six with 1:18 left in the game. There wasn't much running in this game, though Robert Turbin had more success than Marshawn Lynch.

Stevan Ridley dropped a pass early in this one, but still wound up with 16 carries and a reception. Still, Brandon Bolden and Danny Woodhead were involved and ran more successfully.

C.J. Spiller is a top-five real-life back and should be top 12-15 in fantasy even if Fred Jackson stays healthy and gets his full cut of the time share.

In the 11th round of the 12-team, short bench, 1-QB Yahoo Friends and Family league, I had a choice between RGIII and Peyton Manning. I felt Manning was probably the sober pick, but it was a shallow 1-QB league, so I took the guy I wanted. Not that I could have gone wrong with either, but you have to love it when your QB runs for a 76-yard TD.

I was wrong to credit the Packers for last year's performance agains the Colts last week, but I was apparently also wrong not to credit them for it against Houston this week. Obviously, last year's team showed up tonight only with a better defense. And Alex Green looks like a upgrade from James Starks. In one day, the Packers went from roughly top-10 to arguably the favorite to win the Super Bowl.

Things to Look for in Week 7

The Seahawks slug it out in San Francisco Thursday night.

The Ravens missing half their defense travel to Houston

RG and the Redskins play the Giants in the Meadowlands

The Packers head to St. Louis after a breakout win in Houston

Beating the Book

Titans +3 at Bills

I rarely use the same team twice in a row as the market usually adjusts, but in this case it hasn't. The Titans were lucky to beat a depleted Steelers team at home off a short week, but they're still among the league's doormats in my opinion. The Bills looked bad against the Pats for a half and the 49ers for a whole, but overall, they're still a league-average type team. Back Buffalo.

Bills 27 - 17

Last week we won with the Bills to put us at 3-3 on the season in this forum and 46-43-2 overall. Last year we went 10-7 in this forum and 124-125 overall. Over the last five years we've gone 50-34 in this space. You can read the full Week 6 column here.

Surviving Week 7

Last week, I faded the Falcons which given the pot-odds in what turned out to be a coin-flip game was the right call even though they won on a long field goal. Unfortunately, I took the Steelers on Thursday night which was the wrong call, and lost two of my three remaining pools (I took the Dolphins in the other one). Most of the other teams won except the Cardinals, so in the end, most of the people who were alive last week are still around. Let's take a look at this week's slate.

Team Opponent %Taken Vegas ML** Vegas Odds
PATRIOTS Jets 38.10% 525 84.00
49ERS Seahawks 13.30% 300 75.00
Packers Rams 12.40% 230 69.70
VIKINGS Cardinals 9.10% 255 71.83
RAIDERS Jaguars 8.50% 200 66.67
GIANTS Redskins 4.10% 255 71.83
BEARS Lions 3.60% 240 70.59
COLTS Browns 2.70% 160 61.54
TEXANS Ravens 2.30% 280 73.68
BILLS Titans 2.20% 170 62.96

Home Team in CAPS
* according to OfficeFootballPools.com
** average of the two moneylines

At first glance, the Patriots are the easy case far less likely to lose than the No. 2 team, the 49ers and 38 percent isn't a huge number. If you've used the Pats, it gets dicier. I think the 49ers are the second best play, followed by the Texans at home against the banged-up Ravens. The Vikings would probably be my fourth choice and possibly even third, ahead of Baltimore. I'll finalize the entire order on Wednesday night.

Follow Chris on Twitter at @Chris_Liss